Research on Traffic Congestion Warning Method Based on Data Modeling
Chunyao Zhong, Qunlong Huang, Zeping An, Shigui Luo, Hu Meng, Yunfeng Wei · 2023
A congestion warning method based on changes in the slope of traffic flow basic maps is proposed to address the research gap in which existing research cannot effectively warn of congestion using interpretable methods in the absence of high-quality observation data. Generally speaking, highway traffic congestion is caused by actual traffic flow exceeding the road capacity. When congestion occurs, there will be significant changes in traffic flow and speed. This article proposes a congestion warning framework (based on the “data+model” framework), which is based on the fact that during the duration before and after congestion occurs, traffic increases sharply and speed decreases sharply. In order to find a stable and reliable warning point, this article first determines the range of speed and flow values before and after congestion by observing actual traffic flow data. At the same time, through data fitting, select an appropriate traffic flow model to describe the basic relationship between the traffic parameters of the selected road section. Then, a congestion warning model based on the slope change of the basic graph was proposed. Finally, within the preliminary warning range determined by historical data, the proposed model is used to identify early warning points and achieve congestion warning. Through verification of actual highway traffic flow data, it was found that the accuracy rate of early warning is as high as 94.44%.